Week 1B: February 1st, 2023
Working with: ml5 image classification
Additional layers to add to image/object classifiers
-
text, voice, emojis for detected objects. accessorizing with p5 elements
-
game elements
ex. have classifier identify 3 objects
If “word x” is within that list of identified objects/array, then “I see X” > “I found X”
-
translator API
Need to set up API Key though.
Inspiration Examples
- identifies object but also tells you what it rhymes with
- click detected object in image and it’ll play a sound (i.e., animal images, corresponding animal sounds). good for children games
My Own Ideas
- have a computer play charades with you. Given a word and instead of emoji scavenger hunt where you find something, you can make the object with your body like charades and the computer has to detect the object made
- or create a speed game with numbers and rounds. Set number of rounds like 10 rounds, try to beat fastest time. Given random # and have to show same number of fingers
- rock paper scissors but have to find object to have it identified to be a win
- But then they could just have it on hand and it’s not as fun. similar concept but instead of following rock paper scissor logic as win condition, have to find object bigger than something prompted by computer.
- game random prompts. find something [random color], something [bigger / smaller] than [random object/emoji], that starts with letter[random letter]
- not sure if mobile net (coco database) has this info
- teach ASL through gestures in a game. sequences of gestures you memorize for ASL phrases
- might be better for gesture related models instead of image/object recognition
My p5 Sketch